Pacific Science (1983), vol. 37, no. 4 © 1984 by the University of Press. All rights reserved

Regeneration of Southern Rata ( umbellata) and Kamahi ( racemosa) in Areas of Dieback 1

R. B. ALLEN2 and A. B. ROSE2

ABSTRACT: During the 1950s, dieback of southern rata (Metrosideros umbel/ata)umbellata) and kamahi (Weinmannia racemosa) was documented in the conifer/ hardwood forests of the Kokatahi and Fox catchments in Westland, . Thirty years later, southern rata was usually absent as a live in dieback stands. Kamahi, although absent from the canopy in the Kokatahi, had partially recovered at Fox, where it was still a dominant canopy species. Regeneration studies on a range of sites indicate that in the short term, kamahi and Quintinia acutifolia will become the structurally dominant canopy species in many of the dieback stands, and southern rata will at best be a minor component.

DIEBACK OF INDIGENOUS FORESTS is a wide­ gitude, 43°30' S latitude. Both areas are spread phenomenon in New Zealand; it oc­ within the West Coast "beech gap," which ex­ curs in many ofthe important genera, such as tends in latitude from approximately 42°50' S Eodocarpus, , Metrosideros, and to approximately 43°40' S.. Between these Weinmannia. Dieback of southern rata (M. latitudes, Nothofagus species are absent and umbel/ata)3umbellata)3 and kamahi (W. racemosa) is a the dominant canopy hardwood species are feature of the steepland conifer/hardwood southern rata, kamahi, and Quintinia acuti­ forests of Westland, where these species are folia. abundant, and in many other areas where Soils in this area are derived from fractured these species occur. It is generally believed schist along the alpine fault. They are that dieback of southern rata and kamahi is typically skeletal on steeper slopes, with principally due to browsing by the introduced increasing development on lesser slopes and brush-tailed possum (Trichosurus vulpecula podzol formation on gentle slopes. As phy­ Kerr.). The importance of other factors was siography is the result offluvial modification considered by C. G. R. Chavasse (1955, of formerly glaciated valleys, the forested unpublished) and more recently by Veblen slopes are steep, often greater than 30°, and and Stewart (1982) who support the hypo­ are subject to landslides. Such landslides are thesis that much of the apparently excessive usually in the form of debris slides or debris tree mortality is related to natural stand avalanches. Climate is characterized by high dynamics. annual rainfall and high humidity. Extensive The mortality and regeneration patterns in northerly airflows are usually accompanied dieback stands are described for two areas: by prolonged rainfall that may reach torren­ the Kokatahi catchment, at approximately tial intensities. Studies in adjacent areas 171 °10' E longitude, 42°55' S latitude, a tribu­ suggest rainfall reaches a maximum just east tary of the Hokitika River; and the Fox ofthe alpine fault, where it exceeds 1O,000mm catchment, at approximately 1700 00'E lon- per annum. Much of this area receives over 6000mm per annum (Griffiths and Mc­ I N.Z.F.S. No. I684/0DC 182.2. Manuscript accepted Saveney 1982). 5 October 1983. The introduction of browsing animals has 2 Protection Forestry Division, Forest Research Insti­ tute, P.O. Box 31-011, , New Zealand. caused modification to these forests. Brush­ 3Nomenclature follows Allan (1961) unless stated tailed possums have been present in both the otherwise. Kokatahi and Fox catchments for over 40 yr. 433 434 PACIFIC SCIENCE, Volume 37, October 1983

Chamois (Rupicapra rupicapra L.) occur in kamahi and Q. acutifolia were abundant, but both areas. Red deer (Cervus elaphus L.) have occurred mainly on logs. Very few seedlings been present in the Kokatahi for over 40 yr, or saplings of southern rata were present. but are rare at Fox. Numbers have declined Intact stands of small-diameter oc­ markedly in the Kokatahi since the 1950s. curred on the steeper slopes forming the As a consequence of dieback, major escarpments of the Kokatahi River. These changes occur in the structure and composi­ contained southern rata, kamahi, and Q. tion of the affected stands. This paper ex­ acutifolia in the canopy. amines the changes that have resultedfrom dieback in stands dominated by southern rata Mid-Altitude Forests (450-650 m) and kamahi, and the resulting consequences Southern rata, kamahi, and Quintinia for maintenance ofthe structurally dominant acuiifolia were the dominant canopy species canopy species. The forests in each study area in stands before dieback occurred. Broadleaf are first considered in altitudinal classes based (Griselinia littoralis) became more abundant on the canopy species dominant before die­ with increasing altitude. back, followed by general comments on the It is in this zone that mortality has been area. These preliminary results are based on most extensive and complete. Figure 1· pre­ observations and some plot data. sents diameter size class distributions ofthe canopy species from a quadrat established in a representative stand in 1982. Within this KOKATAHICATCHMENT quadrat, all southern rata stems were dead. Dieback in the Kokatahi first became ap­ This included small single-stemmed trees, the patehtin the lat~ 19408. The-mbstex:tensive occasiortalffluchlatgeritH:lividulfl; -and some areas ofdieback in Westland were considered multistemmed trees. One etiolated seedling to be those occurring on the north bank ofthe was the sole representative of southern rata Kokatahi River (J. M. Hoy 1955, unpub­ on 180 m2 ofseedling plots measured on logs, lished). On some ridges between 300 and on tree fern trunks, and onthe ground within 700 m altitude, it was estimated that 75% of the above quadrat. Most of the remaining the original canopy, including southern rata dead stems were kamahi or Quintinia acuti­ and kamahi,was dead (L. W. Boot 1955, folia. There were no live stems ofkamahi, but unpublished). Areas ofup to tens of hectares Q. acutifolia was abundant in the smaller from the valley bottom to the upper al­ diameter size classes. Thedense tier was titudinallimit of southern rata were affected. rich in species, of which Pseudowintera colorata and Carpodetus serratus were the Low-Altitude Forests (300-450m) most abundant subcanopy species, and Q. acutifolia was the dominant canopy species The hardwood canopy species dominant (Figure 2). Many tree ferns· were present, par­ before dieback occurred were southern rata, ticularly Alsophilasmithii (Hook. f.) Tryon. kamahi, and Quintinia acutifolia. The coni­ Seedlings. of kamahi and Q. acutifolia were fers rimu () and miro growing terrestrially as well as epiphytically (Podocarpus ferrugineus) were also present, on logs and tree fern trunks. rimu being limited to low-altitude stable sites. Dieback resulted in the loss of southern High-Altitude Forests (650-750 m) rata, kamahi, and much of the Quintinia acutifolia from the canopy. The podocarps Southern rata and broadleaf trees formed remained healthy. In stands affected by die­ an open-canopied forest, often on slopes of back in the canopy, the more abundant sub­ less than 10°. canopy species were Carpodetus serratus and After dieback, all the southern rata trees Pseudowinteracolorata. Saplings and seed­ and many of the broadleaf trees died (Figure lings of podocarps were present, particularly 3). The only live trees in the canopy were near parent trees. Seedlings and saplings of broadleaf, many of which had only partial

I#'M Regeneration of Southern Rata and Kamahi-ALLEN AND ROSE 435

Dead Stems 100 Metrosideros Others 80 umbel/ata n-79 n -24 60

40

20 Ul E 0 G~I b/l~J ...QI 5- l C/) 14 134 304 -0 QI Live Stems ~ 100 Others ...c Quintinia Griselinia KOKATAHI 1 QI 80 acutifolia litt()ralis n-13 ~ n-625 n-46 Altitude: 500m ~ 60 Aspect: 2500 0 40 Slope: 25 Size: 54m x 36m 20 0 S S 5- 14

FIGURE 1. Percentage of live and relatively intact dead stems for canopy species in diameter size classes: n = number of stems> 104m tall; S = saplings, for stems < 5cm d.b.h.o.b. (diameter at breast height over bark) but > 104m tall; and then in lOcm d.b.h.o.b. size classes. crowns. The most abundant species forming Annual ring counts from disks and increment the shrub tier were Pseudowintera colorata, cores suggest that many of the subcanopy Coprosma ciliata, and Myrsine divaricata. trees and were present before dieback. Broadleaf and many subcanopy species were Most ofthe Q. acutifolia had been established regenerating on the bases of dead trees and after dieback, but individuals that predated fallen logs. However, seedlings of southern the dieback showed more rapid growth over rata were rare. the last 30 yr. Quintinia acutifolia was flower­ Thirty-six 400 m2 quadrats established in ing at ages of less than 20 yr, and because of 1979 over a range offorested sites throughout its abundance in the developing stands was in the Kokatahi catchment showed little ad­ a good position to occupy vacant sites. vance growth of southern rata in open- or The dieback stands described in this study closed-canopied stands, although seedlings were of the earliest described mortality. Die­ and saplings of kamahi and Quintinia acuti­ back in southern rata and kamahi has oc­ folia were relatively abundant. Species highly curred at other times, and extensive areas of palatable to red deer were present within the recent dieback were present on the south side browse tier, indicating that red deer were of the Kokatahi catchment. Some of these unlikely to be limiting the regeneration of trees were not yet dead, although most ofthe southern rata, although this may not always foliage had been lost. have been so. An earlier report considered On recent landslides within areas of die­ that red deer checked the establishment of back, regeneration of kamahi and Quintinia southern rata and kamahi in dieback stands acutifolia was usually dense. Regeneration of in the Kokatahi catchment (Holloway 1959). southern rata was more abundant on land- 436 PACIFIC SCIENCE, Volume 37, October 1983

FIGURE 2. Dead southern rata, kamahi, and Quintinia acutifolia trees in a stand at 500 m altitude. The dominant species in the shrub tier are Q. acutifolia and Pseudowintera colorata. slides near intact forest. From field observa­ FOX CATCHMENT tions, few southern rata seedlings were present on recent landslide surfaces greater Photographs taken of the Fox glacier over than 300 m from seed sources. Browse on the period 1920-1940 show oblique views of southern rata and kamahi seedlings present the forests in which dieback was not apparent. on landslides was attributed to brush-tailed Early reports on the vegetation did not men­ possums. Important colonizing species on tion mortality. However, dead crowns in the these primary sites were Carmichaelia c.f. canopy on the south side ofthe valley can be grandiflora, Olearia avicenniaefolia, and (at seen in photographs taken in the late 1940s higher altitudes) Dracophyllum longifolium. and early 1950s. Death of the canopy trees,

21LIJJ!l2Jt!&WWW_21LIJJ!l2Jt!&WWWW Regeneration of Southern Rata and Kamahi-ALLEN AND ROSE 437

FIGURE 3. A stand ofdead southern rata trees emergent through a dense shrub tier at 700 m altitude. Live broad­ leaf trees remain in the canopy. mainly kamahi and some southern rata, was Quintinia acutifolia, Hall's totara (Podocarpus described by Chavasse (1955, unpublished) as hal/b), and miro. spectacular, particularly on steeper slopes. In stands affected by dieback in the 1950s, Open-canopied forest developed where die­ virtually all the southern rata and much of back took place. the Hall's totara had died, leaving live trees of kamahi and Quintinia acutifolia (Figure 4). From a quadrat established in 1982 in a Low-Altitude Forests (300-600 m) representative stand, diameter size class dis­ Important canopy species before dieback tributions for the canopy species are pre­ included southern rata, kamahi, broadleaf, sented in Figure 5. In this stand, all southern 438 PACIFIC SCIENCE, Volume 37, October 1983

t Jli .,~~ . ..

FIGURE 4. Profile of a low-altitude dieback stand with live kamahi and Quintinia acutifolia trees; the dead trees are southern rata and Hall's totara. The dominant shrub tier species are Coprosma foetidissima and Pseudopanax simplex. Regeneration of Southern Rata and Kamahi-ALLEN AND ROSE 439

FOX 1 Dead Stems Altitude: 530m 100 Melrosideros Aspect: 35° 80 umbel/ala Slope: 250 n=17 Size: 12m x 18m 60

40

(Jl 20 E 4l 0 ---'-1h-~,.-t""J"'"""'-r'~~~Lr".u...-~~L..,-,/ 1:_1 C/)- 1 204 -0 100 4l Podocarpus Others 01 III 80 hallii n= 18 C n=16 -4l (J 60 ~ 4l a. 40 20

20 o

FIGURE 5. Percentage of live and relatively intact dead stems for canopy species in diameter size classes: n = number ofstems > 104m tall; S = saplings, for stems <5cmd.b.h.o.b. but> 104m tall; and then in lOcm d.b.h.o.b. size classe~ 440 PACIFIC SCIENCE, Volume 37, October 1983

rata stems-over a wide diameter range­ were few dead individuals ofthe other canopy were dead, except for one healthy individual species. Regeneration of broadleaf was with a diameter of approximately 40 cm and abundant, but there were very few seedlings five saplings. This stand was relatively rich in of southern rata or kamahi. canopy species, kamahi being the most abun­ Usually, live trees ofsouthern rata were not dant. Much ofthe kamahi had died back, but present in dieback stands within its altitudinal had recovered by sprouting epicormic shoots. range. However, there were intact stands con­ As a result, the intensity and extent of mor­ taining southern rata within the general die­ tality was not as great as photographs of the back area. In the low- and high-altitude dieback taken in the 1950s would indicate. forests, slopes were steep and the topography The dominant subcanopy species in these dissected, resulting in a mosaic ofstand types, stands were Pseudopanax simplex (Forst. f.) some of which were not subject to dieback. Philipson and Coprosma foetidissima. Sap­ A large number of recently dead individuals lings and seedlings of kamahi, Q. acutifolia, of kamahi, Hall's totara, and Libocedrus broadleaf, and miro were present, usually on bidwillii occurred in dieback stands. The logs. The few southern rata seedlings were mortality took place over a short time period restricted to raised surfaces. (less than 10 yr) and has not spread into new areas. Much of the kamahi and Quintinia Mid-Altitude Forests (600-800 m) acutifolia were present as small trees and saplings. From annual ring counts of disks Southern rata, kamahi, and broadleaf were and increment cores, many of these indi­ the dominant canopy species before the viduals were postdieback. This contrasted standswere- affected byoie15ack. S6n1e stands with the subcanopyspecies, which were large­ were dominated entirely by southern rata. ly present before the dieback. After dieback, mortality was most com­ Surfaces exposed by landslides over the plete in stands with a high southern rata last 20 yr within areas of dieback, often with component, because nearly all southern rata logs of the former forest present, had abun­ trees died. Again, much of the kamahi re­ dant regeneration of southern rata, kamahi, covered. Carpodetus serratus was an impor­ and Quintinia acutifolia. For southern rata, tant subcanopy species on talus, with this contrasted with the successions de­ Pseudowintera colorata and Pseudopanax scribed on stable sites. On landslide surfaces, simplex important on stable areas. Broadleaf height growth of southern rata was often at and kamahi were regenerating vigorously, least equal to other woody species. As the with many saplings present and an abundance upper altitudinal limit of a species was of seedlings on logs. Southern rata regenera­ reached, it became less abundant on recent tion was sparse and restricted to log sites. landslides. However, subalpine scrub species (for example, and D. longifolium) colonized landslides well below High-Altitude Forests (800-950m) their lower altitudinal limit in mature com­ Southern rata, kamahi, and broadleaf munities to compete with seedlings ofcanopy were present in stands later affected by die­ species. The shrubs Olearia avicenniaefolia back, along with the conifer Libocedrus bid­ and Coprosma rugosa were also early colo­ willii. With increasing altitude, the subalpine nizers of landslides. The tussock Chionochloa scrub species Dracophyllum longifolium and conspicua Zotov preferred moist areas and D. traversiiwere more important components gullies where it limited the establishment of of the forest canopy. canopy speCIes. Because ofgreater species richness, canopy removal after dieback has not been as com­ plete as in the mid-altitude forests. In dieback DISCUSSION stands, dead southern rata trees were found as widely spaced individuals. Some Liboce­ In the two -study areas, the extent and drus bidwillii trees were also dead, but there intensity of canopy removal after dieback

~il..l(t&dif(& , 5'6 21 ; d WWiiM 9 ' 1fM*1fM" ; me! tffitm' Regeneration of Southern Rata and Kamahi-ALLEN AND ROSE 441

was largely dependent on the spatial pattern changes. Kamahi and Quintinia acutifolia of stands susceptible to dieback and the seedlings are shade-tolerant compared with species richness of the canopy. Apparently southern rata and can develop under rela­ even-aged stands, usually with diameters of tively intact canopies, as well as in canopy less than 30 cm and often originating on gaps. In old stands on stable sites, southern landslide surfaces, remained intact. These rata can maintain a presence by regenerating stands, which contained southern rata, in canopy gaps, often on raised surfaces kamahi, and Quintinia acutifolia, were more (Stewart and Veblen 1982, Wardle 1971). frequent at Fox, although they were still only This intermittent regeneration results in the a small part of the total area. Trees within development of mixed-age stands. Southern these stands were possibly vigorous because rata and kamahi are opportunistic species they occurred on recent mineral surfaces and that can establish themselves on landslide were relatively young. Before dieback oc­ scars (Stewart and Veblen 1982, Wardle curred, the forests in the Kokatahi were less 1980), as does Q. acutifolia. On these sites, rich in canopy species than those in the Fox even-aged stands develop, particularly where catchment. Additional factors led to a more slopes are steep (Stewart and Veblen 1982, complete dieback in the Kokatahi; the Wardle 1980). Southern rata (Wardle 1971), kamahi trees did not recover as they did at kamahi (Wardle 1966), and Q. acutifolia all Fox and the mortality of southern rata and produce small, light wind-dispersed seeds, kamahi spread into other parts of the catch­ an advantage in colonizing areas without a ment. This pattern parallels the spread of seed source. Seeding is unlikely to be limiting brush-tailed possums (1. L. James et al. 1973, for southern rata, because some flowers are unpublished).- produced each year, with good flowering There were other floristic differences be­ every 3-4 yr (Wardle 1971), which can prob­ tween the two areas studied. In the Kokatahi, ably be related to frequency of good seed the upper altitudinal limit of many species years. It is not likely that the small seeds of was lower than at Fox; for example, southern any of these species remain viable for long rata was found to an altitude of 750 m in the under natural conditions. Vegetative propa­ Kokatahi, but up to 950 m at Fox. Regenera­ gation of canopy species was not important tion of Quintinia acutifolia was less abundant in these dieback stands, except for the epi­ at Fox, which probably reflects its lesser im­ carmic recovery ofkamahi noted at Fox. portance in the canopy. Understories were The wide diameter range ofdead southern dense in both areas, although tree ferns and rata stems found on the quadrats indicates Pseudowintera colorata were more abundant that these were mixed-age stands (cf. Cole­ in the Kokatahi. Pseudopanax simplex and man, Gillman, and Green 1980). This proba­ Carpodetus serratus, species palatable to deer, bly also applies to kamahi. In general, the were more abundant at Fox. This can be regenerative process that established these related to the absence of deer at Fox, but it stands would have been individual tree or is also likely that some of these species were group replacement requiring relatively stable more abundant because they are early succes­ landforms, such as ridges. This conclusion sional species that prefer mineral soils. conflicts with previous studies in that it does The change in physiognomy of the forests not support the idea that dieback occurs in caused by dieback has important conse­ synchronously senescent stands established quences. The density of the understory in­ after large-scale disturbances (Chavasse 1955, creases, even though the understory of forest unpublished; Veblen and Stewart 1982). in the schist zone is already naturally diverse. Hoy's (1955, unpublished) contention that The seed source of species that die back is mortality of southern rata and kamahi is locally removed. Mineral surfaces can be heaviest on steep slopes is merely a reflection formed by landslides in areas of dieback, of where most southern rata and kamahi sometimes as the result ofthe death ofcanopy occur. Dieback is present on a range of sites, trees. including gentle slopes. Individual species react differently to such The effect of dieback is to create large 442 PACIFIC SCIENCE, Volume 37, October 1983 canopy gaps over short time periods. The as the structurally dominant canopy species. gradual opening up of the canopy and the Extensive even-aged stands of these species presence of a dense shrub tier favors the can be expected to develop in most dieback more shade-tolerant canopy species. Where stands. On many sites, these stands will the extent of dieback increases, as in the include other canopy species such as broad­ Kokatahi, a point is reached where the seed leaf, rimu, miro, or Hall's totara. available for regeneration is limiting. This is particularly important for southern rata, which has limited sites for establishment. LITERATURE CITED Standing dead trees are not as suited for the establishment of seedlings as fallen logs. Al­ ALLAN, H. H. 1961. Flora of New Zealand. though southern rata seedlings are compara­ Vol. I. Govt. Printer, Wellington, New tively abundant on landslide surfaces, at Zealand. present these are only a small part of the COLEMAN, J. D., A. GILLMAN, and W. Q. total surface area, even at Fox. In larger GREEN. 1980. Forest patterns and possum areas of dieback, availability of seed appears densities within podocarp-/mixed hard­ to become limiting as there are few southern forests on Mt. Bryan O'Lynn, West­ rata seedlings, even on the most suitable sites. land. New Zealand J. Ecol. 3: 69-84. It is concluded that this species will be neither GRIFFITHS, G. A., and M. J. MCSAVENEY. as abundant nor as extensive in the post­ 1982. Distribution of mean annual pre­ dieback stands as it was before dieback. In cipitation across some steepland regions some areas like the Kokatahi, where dieback of New Zealand. Christchurch Water and is extensive,-if will remain virtually aHSefit Soil Sci<::nce CentreR~ept. WS648; Ministry as a canopy tree for at least one generation ofWorks and Development, Christchurch. of the replacement canopy. Southern rata is HOLLOWAY,J. T.1959. Noxiousanimalprob­ a relatively long-lived species in these forests, lems ofthe alpine watersheds. with slow recruitment rates. Provided seed New Zealand Sci. Rev. 17(2):4-10. sources are near, southern rata will be es­ STEWART, G. H., and T. T. VEBLEN. 1982. tablished as sites become available, but it Regeneration patterns in southern rata will take a number of generations before it ()-kamahi (Wein­ can possibly establish its former abundance. mannia racemosa) forest in central West­ At Kelly's Creek, central Westland, Stewart land, New Zealand. New Zealand J. Bot. and Veblen (1982) described vigorous re­ 20:55-72. generation of the main canopy species, in­ VEBLEN, T. T., and G. H. STEWART. 1982. cluding southern rata, after "massive" mor­ The effects of introduced wild animals on tality. By comparison, the mortality of New Zealand forests. Ann. Assoc. Amer. southern rata in dieback stands in the Geogr. 72(3): 372-397. Kokatahi and Fox catchments has been more WARDLE, P. 1966. Biological flora of New complete and extensive, and the density of Zealand. I. Weinmannia racemosa Linn. regeneration is much lower. F. () kamahi. New Zealand Kamahi and Quintinia acutifolia were J. Bot. 4: 114-31. found to regenerate successfully under a ---. 1971. Biological flora of New Zea­ greater range of conditions than southern land. 6. Metrosideros umbellata Cav. [Syn. rata. Although removal ofthese species from M. lucida (Forst. f.) A. Rich.] () the canopy has not been as complete as southern rata. New Zealand J. Bot. 9: 645­ southern rata, it is considered that this 671. success is due to their greater competitive ---.1980. Primary succession in Westland ability. The successions after dieback in the National Park and its vicinity, New Zea­ areas studied favor kamahi and Q. acutifolia land. New Zealand J. Bot. 18:221-232.

L&WJZZiZ!LlIUifillZP. ; WI